|
542
|
1 """
|
|
|
2 Generate Reaction Activity Scores (RAS) from a gene expression dataset and GPR rules.
|
|
|
3
|
|
|
4 The script reads a tabular dataset (genes x samples) and a rules file (GPRs),
|
|
|
5 computes RAS per reaction for each sample/cell line, and writes a tabular output.
|
|
|
6 """
|
|
|
7 from __future__ import division
|
|
|
8 import sys
|
|
|
9 import argparse
|
|
|
10 import pandas as pd
|
|
|
11 import numpy as np
|
|
|
12 try:
|
|
|
13 from .utils import general_utils as utils
|
|
|
14 except:
|
|
|
15 import utils.general_utils as utils
|
|
|
16 from typing import List, Dict
|
|
|
17 import ast
|
|
|
18
|
|
|
19 # Optional imports for AnnData mode (not used in ras_generator.py)
|
|
|
20 try:
|
|
|
21 from progressbar import ProgressBar, Bar, Percentage
|
|
|
22 from scanpy import AnnData
|
|
|
23 from cobra.flux_analysis.variability import find_essential_reactions, find_essential_genes
|
|
|
24 except ImportError:
|
|
|
25 # These are only needed for AnnData mode, not for ras_generator.py
|
|
|
26 pass
|
|
|
27
|
|
|
28 ERRORS = []
|
|
|
29 ########################## argparse ##########################################
|
|
|
30 ARGS :argparse.Namespace
|
|
|
31 def process_args(args:List[str] = None) -> argparse.Namespace:
|
|
|
32 """
|
|
|
33 Processes command-line arguments.
|
|
|
34
|
|
|
35 Args:
|
|
|
36 args (list): List of command-line arguments.
|
|
|
37
|
|
|
38 Returns:
|
|
|
39 Namespace: An object containing parsed arguments.
|
|
|
40 """
|
|
|
41 parser = argparse.ArgumentParser(
|
|
|
42 usage = '%(prog)s [options]',
|
|
|
43 description = "process some value's genes to create a comparison's map.")
|
|
|
44
|
|
|
45 parser.add_argument("-rl", "--model_upload", type = str,
|
|
|
46 help = "path to input file containing the rules")
|
|
|
47
|
|
|
48 parser.add_argument("-rn", "--model_upload_name", type = str, help = "custom rules name")
|
|
|
49 # Galaxy converts files into .dat, this helps infer the original extension when needed.
|
|
|
50
|
|
|
51 parser.add_argument(
|
|
|
52 '-n', '--none',
|
|
|
53 type = utils.Bool("none"), default = True,
|
|
|
54 help = 'compute Nan values')
|
|
|
55
|
|
|
56 parser.add_argument(
|
|
|
57 '-td', '--tool_dir',
|
|
|
58 type = str,
|
|
|
59 default = os.path.dirname(os.path.abspath(__file__)),
|
|
|
60 help = 'your tool directory (default: auto-detected package location)')
|
|
|
61
|
|
|
62 parser.add_argument(
|
|
|
63 '-ol', '--out_log',
|
|
|
64 type = str,
|
|
|
65 help = "Output log")
|
|
|
66
|
|
|
67 parser.add_argument(
|
|
|
68 '-in', '--input',
|
|
|
69 type = str,
|
|
|
70 help = 'input dataset')
|
|
|
71
|
|
|
72 parser.add_argument(
|
|
|
73 '-ra', '--ras_output',
|
|
|
74 type = str,
|
|
|
75 required = True, help = 'ras output')
|
|
|
76
|
|
|
77
|
|
|
78 return parser.parse_args(args)
|
|
|
79
|
|
|
80 ############################ dataset input ####################################
|
|
|
81 def read_dataset(data :str, name :str) -> pd.DataFrame:
|
|
|
82 """
|
|
|
83 Read a dataset from a CSV file and return it as a pandas DataFrame.
|
|
|
84
|
|
|
85 Args:
|
|
|
86 data (str): Path to the CSV file containing the dataset.
|
|
|
87 name (str): Name of the dataset, used in error messages.
|
|
|
88
|
|
|
89 Returns:
|
|
|
90 pandas.DataFrame: DataFrame containing the dataset.
|
|
|
91
|
|
|
92 Raises:
|
|
|
93 pd.errors.EmptyDataError: If the CSV file is empty.
|
|
|
94 sys.exit: If the CSV file has the wrong format, the execution is aborted.
|
|
|
95 """
|
|
|
96 try:
|
|
|
97 dataset = pd.read_csv(data, sep = '\t', header = 0, engine='python', index_col=0)
|
|
|
98 dataset = dataset.astype(float)
|
|
|
99 except pd.errors.EmptyDataError:
|
|
|
100 sys.exit('Execution aborted: wrong file format of ' + name + '\n')
|
|
|
101 if len(dataset.columns) < 2:
|
|
|
102 sys.exit('Execution aborted: wrong file format of ' + name + '\n')
|
|
|
103 return dataset
|
|
|
104
|
|
|
105
|
|
|
106 def load_custom_rules() -> Dict[str,str]:
|
|
|
107 """
|
|
|
108 Opens custom rules file and extracts the rules. If the file is in .csv format an additional parsing step will be
|
|
|
109 performed, significantly impacting the runtime.
|
|
|
110
|
|
|
111 Returns:
|
|
|
112 Dict[str, ruleUtils.OpList] : dict mapping reaction IDs to rules.
|
|
|
113 """
|
|
|
114 datFilePath = utils.FilePath.fromStrPath(ARGS.model_upload) # actual file, stored in Galaxy as a .dat
|
|
|
115
|
|
|
116 dict_rule = {}
|
|
|
117
|
|
|
118 try:
|
|
|
119 rows = utils.readCsv(datFilePath, delimiter = "\t", skipHeader=False)
|
|
|
120 if len(rows) <= 1:
|
|
|
121 raise ValueError("Model tabular with 1 column is not supported.")
|
|
|
122
|
|
|
123 if not rows:
|
|
|
124 raise ValueError("Model tabular is file is empty.")
|
|
|
125
|
|
|
126 id_idx, idx_gpr = utils.findIdxByName(rows[0], "GPR")
|
|
|
127
|
|
|
128 # First, try using a tab delimiter
|
|
|
129 for line in rows[1:]:
|
|
|
130 if len(line) <= idx_gpr:
|
|
|
131 utils.logWarning(f"Skipping malformed line: {line}", ARGS.out_log)
|
|
|
132 continue
|
|
|
133
|
|
|
134 dict_rule[line[id_idx]] = line[idx_gpr]
|
|
|
135
|
|
|
136 except Exception as e:
|
|
|
137 # If parsing with tabs fails, try comma delimiter
|
|
|
138 try:
|
|
|
139 rows = utils.readCsv(datFilePath, delimiter = ",", skipHeader=False)
|
|
|
140
|
|
|
141 if len(rows) <= 1:
|
|
|
142 raise ValueError("Model tabular with 1 column is not supported.")
|
|
|
143
|
|
|
144 if not rows:
|
|
|
145 raise ValueError("Model tabular is file is empty.")
|
|
|
146
|
|
|
147 id_idx, idx_gpr = utils.findIdxByName(rows[0], "GPR")
|
|
|
148
|
|
|
149 # Try again parsing row content with the GPR column using comma-separated values
|
|
|
150 for line in rows[1:]:
|
|
|
151 if len(line) <= idx_gpr:
|
|
|
152 utils.logWarning(f"Skipping malformed line: {line}", ARGS.out_log)
|
|
|
153 continue
|
|
|
154
|
|
|
155 dict_rule[line[id_idx]] =line[idx_gpr]
|
|
|
156
|
|
|
157 except Exception as e2:
|
|
|
158 raise ValueError(f"Unable to parse rules file. Tried both tab and comma delimiters. Original errors: Tab: {e}, Comma: {e2}")
|
|
|
159
|
|
|
160 if not dict_rule:
|
|
|
161 raise ValueError("No valid rules found in the uploaded file. Please check the file format.")
|
|
|
162 # csv rules need to be parsed, those in a pickle format are taken to be pre-parsed.
|
|
|
163 return dict_rule
|
|
|
164
|
|
|
165
|
|
|
166 """
|
|
|
167 Class to compute the RAS values
|
|
|
168
|
|
|
169 """
|
|
|
170
|
|
|
171 class RAS_computation:
|
|
|
172
|
|
|
173 def __init__(self, adata=None, model=None, dataset=None, gene_rules=None, rules_total_string=None):
|
|
|
174 """
|
|
|
175 Initialize RAS computation with two possible input modes:
|
|
|
176
|
|
|
177 Mode 1 (Original - for sampling_main.py):
|
|
|
178 adata: AnnData object with gene expression (cells × genes)
|
|
|
179 model: COBRApy model object with reactions and GPRs
|
|
|
180
|
|
|
181 Mode 2 (New - for ras_generator.py):
|
|
|
182 dataset: pandas DataFrame with gene expression (genes × samples)
|
|
|
183 gene_rules: dict mapping reaction IDs to GPR strings
|
|
|
184 rules_total_string: list of all gene names in GPRs (for validation)
|
|
|
185 """
|
|
|
186 self._logic_operators = ['and', 'or', '(', ')']
|
|
|
187 self.val_nan = np.nan
|
|
|
188
|
|
|
189 # Determine which mode we're in
|
|
|
190 if adata is not None and model is not None:
|
|
|
191 # Mode 1: AnnData + COBRApy model (original)
|
|
|
192 self._init_from_anndata(adata, model)
|
|
|
193 elif dataset is not None and gene_rules is not None:
|
|
|
194 # Mode 2: DataFrame + rules dict (ras_generator style)
|
|
|
195 self._init_from_dataframe(dataset, gene_rules, rules_total_string)
|
|
|
196 else:
|
|
|
197 raise ValueError(
|
|
|
198 "Invalid initialization. Provide either:\n"
|
|
|
199 " - adata + model (for AnnData input), or\n"
|
|
|
200 " - dataset + gene_rules (for DataFrame input)"
|
|
|
201 )
|
|
|
202
|
|
|
203 def _normalize_gene_name(self, gene_name):
|
|
|
204 """Normalize gene names by replacing special characters."""
|
|
|
205 return gene_name.replace("-", "_").replace(":", "_")
|
|
|
206
|
|
|
207 def _normalize_rule(self, rule):
|
|
|
208 """Normalize GPR rule: lowercase operators, add spaces around parentheses, normalize gene names."""
|
|
|
209 rule = rule.replace("OR", "or").replace("AND", "and")
|
|
|
210 rule = rule.replace("(", "( ").replace(")", " )")
|
|
|
211 # Normalize gene names in the rule
|
|
|
212 tokens = rule.split()
|
|
|
213 normalized_tokens = [token if token in self._logic_operators else self._normalize_gene_name(token) for token in tokens]
|
|
|
214 return " ".join(normalized_tokens)
|
|
|
215
|
|
|
216 def _init_from_anndata(self, adata, model):
|
|
|
217 """Initialize from AnnData and COBRApy model (original mode)."""
|
|
|
218 # Build the dictionary for the GPRs
|
|
|
219 df_reactions = pd.DataFrame(index=[reaction.id for reaction in model.reactions])
|
|
|
220 gene_rules = [self._normalize_rule(reaction.gene_reaction_rule) for reaction in model.reactions]
|
|
|
221 df_reactions['rule'] = gene_rules
|
|
|
222 df_reactions = df_reactions.reset_index()
|
|
|
223 df_reactions = df_reactions.groupby('rule').agg(lambda x: sorted(list(x)))
|
|
|
224
|
|
|
225 self.dict_rule_reactions = df_reactions.to_dict()['index']
|
|
|
226
|
|
|
227 # build useful structures for RAS computation
|
|
|
228 self.model = model
|
|
|
229 self.count_adata = adata.copy()
|
|
|
230
|
|
|
231 # Normalize gene names in both model and dataset
|
|
|
232 model_genes = [self._normalize_gene_name(gene.id) for gene in model.genes]
|
|
|
233 dataset_genes = [self._normalize_gene_name(gene) for gene in self.count_adata.var.index]
|
|
|
234 self.genes = pd.Index(dataset_genes).intersection(model_genes)
|
|
|
235
|
|
|
236 if len(self.genes) == 0:
|
|
|
237 raise ValueError("ERROR: No genes from the count matrix match the metabolic model. Check that gene annotations are consistent between model and dataset.")
|
|
|
238
|
|
|
239 self.cell_ids = list(self.count_adata.obs.index.values)
|
|
|
240 # Get expression data with normalized gene names
|
|
|
241 self.count_df_filtered = self.count_adata.to_df().T
|
|
|
242 self.count_df_filtered.index = [self._normalize_gene_name(g) for g in self.count_df_filtered.index]
|
|
|
243 self.count_df_filtered = self.count_df_filtered.loc[self.genes]
|
|
|
244
|
|
|
245 def _init_from_dataframe(self, dataset, gene_rules, rules_total_string):
|
|
|
246 """Initialize from DataFrame and rules dict (ras_generator mode)."""
|
|
|
247 reactions = list(gene_rules.keys())
|
|
|
248
|
|
|
249 # Build the dictionary for the GPRs
|
|
|
250 df_reactions = pd.DataFrame(index=reactions)
|
|
|
251 gene_rules_list = [self._normalize_rule(gene_rules[reaction_id]) for reaction_id in reactions]
|
|
|
252 df_reactions['rule'] = gene_rules_list
|
|
|
253 df_reactions = df_reactions.reset_index()
|
|
|
254 df_reactions = df_reactions.groupby('rule').agg(lambda x: sorted(list(x)))
|
|
|
255
|
|
|
256 self.dict_rule_reactions = df_reactions.to_dict()['index']
|
|
|
257
|
|
|
258 # build useful structures for RAS computation
|
|
|
259 self.model = None
|
|
|
260 self.count_adata = None
|
|
|
261
|
|
|
262 # Normalize gene names in dataset
|
|
|
263 dataset_normalized = dataset.copy()
|
|
|
264 dataset_normalized.index = [self._normalize_gene_name(g) for g in dataset_normalized.index]
|
|
|
265
|
|
|
266 # Determine which genes are in both dataset and GPRs
|
|
|
267 if rules_total_string is not None:
|
|
|
268 rules_genes = [self._normalize_gene_name(g) for g in rules_total_string]
|
|
|
269 self.genes = dataset_normalized.index.intersection(rules_genes)
|
|
|
270 else:
|
|
|
271 # Extract all genes from rules
|
|
|
272 all_genes_in_rules = set()
|
|
|
273 for rule in gene_rules_list:
|
|
|
274 tokens = rule.split()
|
|
|
275 for token in tokens:
|
|
|
276 if token not in self._logic_operators:
|
|
|
277 all_genes_in_rules.add(token)
|
|
|
278 self.genes = dataset_normalized.index.intersection(all_genes_in_rules)
|
|
|
279
|
|
|
280 if len(self.genes) == 0:
|
|
|
281 raise ValueError("ERROR: No genes from the count matrix match the metabolic model. Check that gene annotations are consistent between model and dataset.")
|
|
|
282
|
|
|
283 self.cell_ids = list(dataset_normalized.columns)
|
|
|
284 self.count_df_filtered = dataset_normalized.loc[self.genes]
|
|
|
285
|
|
|
286 def compute(self,
|
|
|
287 or_expression=np.sum, # type of operation to do in case of an or expression (sum, max, mean)
|
|
|
288 and_expression=np.min, # type of operation to do in case of an and expression(min, sum)
|
|
|
289 drop_na_rows=False, # if True remove the nan rows of the ras matrix
|
|
|
290 drop_duplicates=False, # if true, remove duplicates rows
|
|
|
291 ignore_nan=True, # if True, ignore NaN values in GPR evaluation (e.g., A or NaN -> A)
|
|
|
292 print_progressbar=True, # if True, print the progress bar
|
|
|
293 add_count_metadata=True, # if True add metadata of cells in the ras adata
|
|
|
294 add_met_metadata=True, # if True add metadata from the metabolic model (gpr and compartments of reactions)
|
|
|
295 add_essential_reactions=False,
|
|
|
296 add_essential_genes=False
|
|
|
297 ):
|
|
|
298
|
|
|
299 self.or_function = or_expression
|
|
|
300 self.and_function = and_expression
|
|
|
301
|
|
|
302 ras_df = np.full((len(self.dict_rule_reactions), len(self.cell_ids)), np.nan)
|
|
|
303 genes_not_mapped = set() # Track genes not in dataset
|
|
|
304
|
|
|
305 if print_progressbar:
|
|
|
306 pbar = ProgressBar(widgets=[Percentage(), Bar()], maxval=len(self.dict_rule_reactions)).start()
|
|
|
307
|
|
|
308 # Process each unique GPR rule
|
|
|
309 for ind, (rule, reaction_ids) in enumerate(self.dict_rule_reactions.items()):
|
|
|
310 if len(rule) == 0:
|
|
|
311 # Empty rule - keep as NaN
|
|
|
312 pass
|
|
|
313 else:
|
|
|
314 # Extract genes from rule
|
|
|
315 rule_genes = [token for token in rule.split() if token not in self._logic_operators]
|
|
|
316 rule_genes_unique = list(set(rule_genes))
|
|
|
317
|
|
|
318 # Which genes are in the dataset?
|
|
|
319 genes_present = [g for g in rule_genes_unique if g in self.genes]
|
|
|
320 genes_missing = [g for g in rule_genes_unique if g not in self.genes]
|
|
|
321
|
|
|
322 if genes_missing:
|
|
|
323 genes_not_mapped.update(genes_missing)
|
|
|
324
|
|
|
325 if len(genes_present) == 0:
|
|
|
326 # No genes in dataset - keep as NaN
|
|
|
327 pass
|
|
|
328 elif len(genes_missing) > 0 and not ignore_nan:
|
|
|
329 # Some genes missing and we don't ignore NaN - set to NaN
|
|
|
330 pass
|
|
|
331 else:
|
|
|
332 # Evaluate the GPR expression using AST
|
|
|
333 # For single gene, AST handles it fine: ast.parse("GENE_A") works
|
|
|
334 # more genes in the formula
|
|
|
335 check_only_and=("and" in rule and "or" not in rule) #only and
|
|
|
336 check_only_or=("or" in rule and "and" not in rule) #only or
|
|
|
337 if check_only_and or check_only_or:
|
|
|
338 #or/and sequence
|
|
|
339 matrix = self.count_df_filtered.loc[genes_present].values
|
|
|
340 #compute for all cells
|
|
|
341 if check_only_and:
|
|
|
342 ras_df[ind] = self.and_function(matrix, axis=0)
|
|
|
343 else:
|
|
|
344 ras_df[ind] = self.or_function(matrix, axis=0)
|
|
|
345 else:
|
|
|
346 # complex expression (e.g. A or (B and C))
|
|
|
347 data = self.count_df_filtered.loc[genes_present] # dataframe of genes in the GPRs
|
|
|
348 tree = ast.parse(rule, mode="eval").body
|
|
|
349 values_by_cell = [dict(zip(data.index, data[col].values)) for col in data.columns]
|
|
|
350 for j, values in enumerate(values_by_cell):
|
|
|
351 ras_df[ind, j] =self._evaluate_ast(tree, values, self.or_function, self.and_function, ignore_nan)
|
|
|
352
|
|
|
353 if print_progressbar:
|
|
|
354 pbar.update(ind + 1)
|
|
|
355
|
|
|
356 if print_progressbar:
|
|
|
357 pbar.finish()
|
|
|
358
|
|
|
359 # Store genes not mapped for later use
|
|
|
360 self.genes_not_mapped = sorted(genes_not_mapped)
|
|
|
361
|
|
|
362 # create the dataframe of ras (rules x samples)
|
|
|
363 ras_df = pd.DataFrame(data=ras_df, index=range(len(self.dict_rule_reactions)), columns=self.cell_ids)
|
|
|
364 ras_df['Reactions'] = [reaction_ids for rule, reaction_ids in self.dict_rule_reactions.items()]
|
|
|
365
|
|
|
366 reactions_common = pd.DataFrame()
|
|
|
367 reactions_common["Reactions"] = ras_df['Reactions']
|
|
|
368 reactions_common["proof2"] = ras_df['Reactions']
|
|
|
369 reactions_common = reactions_common.explode('Reactions')
|
|
|
370 reactions_common = reactions_common.set_index("Reactions")
|
|
|
371
|
|
|
372 ras_df = ras_df.explode("Reactions")
|
|
|
373 ras_df = ras_df.set_index("Reactions")
|
|
|
374
|
|
|
375 if drop_na_rows:
|
|
|
376 ras_df = ras_df.dropna(how="all")
|
|
|
377
|
|
|
378 if drop_duplicates:
|
|
|
379 ras_df = ras_df.drop_duplicates()
|
|
|
380
|
|
|
381 # If initialized from DataFrame (ras_generator mode), return DataFrame instead of AnnData
|
|
|
382 if self.count_adata is None:
|
|
|
383 return ras_df, self.genes_not_mapped
|
|
|
384
|
|
|
385 # Original AnnData mode: create AnnData structure for RAS
|
|
|
386 ras_adata = AnnData(ras_df.T)
|
|
|
387
|
|
|
388 #add metadata
|
|
|
389 if add_count_metadata:
|
|
|
390 ras_adata.var["common_gprs"] = reactions_common.loc[ras_df.index]
|
|
|
391 ras_adata.var["common_gprs"] = ras_adata.var["common_gprs"].apply(lambda x: ",".join(x))
|
|
|
392 for el in self.count_adata.obs.columns:
|
|
|
393 ras_adata.obs["countmatrix_"+el]=self.count_adata.obs[el]
|
|
|
394
|
|
|
395 if add_met_metadata:
|
|
|
396 if self.model is not None and len(self.model.compartments)>0:
|
|
|
397 ras_adata.var['compartments']=[list(self.model.reactions.get_by_id(reaction).compartments) for reaction in ras_adata.var.index]
|
|
|
398 ras_adata.var['compartments']=ras_adata.var["compartments"].apply(lambda x: ",".join(x))
|
|
|
399
|
|
|
400 if self.model is not None:
|
|
|
401 ras_adata.var['GPR rule'] = [self.model.reactions.get_by_id(reaction).gene_reaction_rule for reaction in ras_adata.var.index]
|
|
|
402
|
|
|
403 if add_essential_reactions:
|
|
|
404 if self.model is not None:
|
|
|
405 essential_reactions=find_essential_reactions(self.model)
|
|
|
406 essential_reactions=[el.id for el in essential_reactions]
|
|
|
407 ras_adata.var['essential reactions']=["yes" if el in essential_reactions else "no" for el in ras_adata.var.index]
|
|
|
408
|
|
|
409 if add_essential_genes:
|
|
|
410 if self.model is not None:
|
|
|
411 essential_genes=find_essential_genes(self.model)
|
|
|
412 essential_genes=[el.id for el in essential_genes]
|
|
|
413 ras_adata.var['essential genes']=[" ".join([gene for gene in genes.split() if gene in essential_genes]) for genes in ras_adata.var["GPR rule"]]
|
|
|
414
|
|
|
415 return ras_adata
|
|
|
416
|
|
|
417 def _evaluate_ast(self, node, values, or_function, and_function, ignore_nan):
|
|
|
418 """
|
|
|
419 Evaluate a boolean expression using AST (Abstract Syntax Tree).
|
|
|
420 Handles all GPR types: single gene, simple (A and B), nested (A or (B and C)).
|
|
|
421
|
|
|
422 Args:
|
|
|
423 node: AST node to evaluate
|
|
|
424 values: Dictionary mapping gene names to their expression values
|
|
|
425 or_function: Function to apply for OR operations
|
|
|
426 and_function: Function to apply for AND operations
|
|
|
427 ignore_nan: If True, ignore None/NaN values (e.g., A or None -> A)
|
|
|
428
|
|
|
429 Returns:
|
|
|
430 Evaluated expression result (float or np.nan)
|
|
|
431 """
|
|
|
432 if isinstance(node, ast.BoolOp):
|
|
|
433 # Boolean operation (and/or)
|
|
|
434 vals = [self._evaluate_ast(v, values, or_function, and_function, ignore_nan) for v in node.values]
|
|
|
435
|
|
|
436 if ignore_nan:
|
|
|
437 # Filter out None/NaN values
|
|
|
438 vals = [v for v in vals if v is not None and not (isinstance(v, float) and np.isnan(v))]
|
|
|
439
|
|
|
440 if not vals:
|
|
|
441 return np.nan
|
|
|
442
|
|
|
443 if isinstance(node.op, ast.Or):
|
|
|
444 return or_function(vals)
|
|
|
445 elif isinstance(node.op, ast.And):
|
|
|
446 return and_function(vals)
|
|
|
447
|
|
|
448 elif isinstance(node, ast.Name):
|
|
|
449 # Variable (gene name)
|
|
|
450 return values.get(node.id, None)
|
|
|
451 elif isinstance(node, ast.Constant):
|
|
|
452 # Constant (shouldn't happen in GPRs, but handle it)
|
|
|
453 return values.get(str(node.value), None)
|
|
|
454 else:
|
|
|
455 raise ValueError(f"Unexpected node type in GPR: {ast.dump(node)}")
|
|
|
456
|
|
|
457
|
|
|
458 # ============================================================================
|
|
|
459 # STANDALONE FUNCTION FOR RAS_GENERATOR COMPATIBILITY
|
|
|
460 # ============================================================================
|
|
|
461
|
|
|
462 def computeRAS(
|
|
|
463 dataset,
|
|
|
464 gene_rules,
|
|
|
465 rules_total_string,
|
|
|
466 or_function=np.sum,
|
|
|
467 and_function=np.min,
|
|
|
468 ignore_nan=True
|
|
|
469 ):
|
|
|
470 """
|
|
|
471 Compute RAS from tabular data and GPR rules (ras_generator.py compatible).
|
|
|
472
|
|
|
473 This is a standalone function that wraps the RAS_computation class
|
|
|
474 to provide the same interface as ras_generator.py.
|
|
|
475
|
|
|
476 Args:
|
|
|
477 dataset: pandas DataFrame with gene expression (genes × samples)
|
|
|
478 gene_rules: dict mapping reaction IDs to GPR strings
|
|
|
479 rules_total_string: list of all gene names in GPRs
|
|
|
480 or_function: function for OR operations (default: np.sum)
|
|
|
481 and_function: function for AND operations (default: np.min)
|
|
|
482 ignore_nan: if True, ignore NaN in GPR evaluation (default: True)
|
|
|
483
|
|
|
484 Returns:
|
|
|
485 tuple: (ras_df, genes_not_mapped)
|
|
|
486 - ras_df: DataFrame with RAS values (reactions × samples)
|
|
|
487 - genes_not_mapped: list of genes in GPRs not found in dataset
|
|
|
488 """
|
|
|
489 # Create RAS computation object in DataFrame mode
|
|
|
490 ras_obj = RAS_computation(
|
|
|
491 dataset=dataset,
|
|
|
492 gene_rules=gene_rules,
|
|
|
493 rules_total_string=rules_total_string
|
|
|
494 )
|
|
|
495
|
|
|
496 # Compute RAS
|
|
|
497 result = ras_obj.compute(
|
|
|
498 or_expression=or_function,
|
|
|
499 and_expression=and_function,
|
|
|
500 ignore_nan=ignore_nan,
|
|
|
501 print_progressbar=False, # No progress bar for ras_generator
|
|
|
502 add_count_metadata=False, # No metadata in DataFrame mode
|
|
|
503 add_met_metadata=False,
|
|
|
504 add_essential_reactions=False,
|
|
|
505 add_essential_genes=False
|
|
|
506 )
|
|
|
507
|
|
|
508 # Result is a tuple (ras_df, genes_not_mapped) in DataFrame mode
|
|
|
509 return result
|
|
|
510
|
|
|
511 def main(args:List[str] = None) -> None:
|
|
|
512 """
|
|
|
513 Initializes everything and sets the program in motion based on the fronted input arguments.
|
|
|
514
|
|
|
515 Returns:
|
|
|
516 None
|
|
|
517 """
|
|
|
518 # get args from frontend (related xml)
|
|
|
519 global ARGS
|
|
|
520 ARGS = process_args(args)
|
|
|
521
|
|
|
522 # read dataset and remove versioning from gene names
|
|
|
523 dataset = read_dataset(ARGS.input, "dataset")
|
|
|
524 orig_gene_list=dataset.index.copy()
|
|
|
525 dataset.index = [str(el.split(".")[0]) for el in dataset.index]
|
|
|
526
|
|
|
527 #load GPR rules
|
|
|
528 rules = load_custom_rules()
|
|
|
529
|
|
|
530 #create a list of all the gpr
|
|
|
531 rules_total_string=""
|
|
|
532 for id,rule in rules.items():
|
|
|
533 rules_total_string+=rule.replace("(","").replace(")","") + " "
|
|
|
534 rules_total_string=list(set(rules_total_string.split(" ")))
|
|
|
535
|
|
|
536 if any(dataset.index.duplicated(keep=False)):
|
|
|
537 genes_duplicates=orig_gene_list[dataset.index.duplicated(keep=False)]
|
|
|
538 genes_duplicates_in_model=[elem for elem in genes_duplicates if elem in rules_total_string]
|
|
|
539
|
|
|
540 if len(genes_duplicates_in_model)>0:#metabolic genes have duplicated entries in the dataset
|
|
|
541 list_str=", ".join(genes_duplicates_in_model)
|
|
|
542 list_genes=f"ERROR: Duplicate entries in the gene dataset present in one or more GPR. The following metabolic genes are duplicated: "+list_str
|
|
|
543 raise ValueError(list_genes)
|
|
|
544 else:
|
|
|
545 list_str=", ".join(genes_duplicates)
|
|
|
546 list_genes=f"INFO: Duplicate entries in the gene dataset. The following genes are duplicated in the dataset but not mentioned in the GPRs: "+list_str
|
|
|
547 utils.logWarning(list_genes,ARGS.out_log)
|
|
|
548
|
|
|
549 #check if nan value must be ignored in the GPR
|
|
|
550 if ARGS.none:
|
|
|
551 # #e.g. (A or nan --> A)
|
|
|
552 ignore_nan = True
|
|
|
553 else:
|
|
|
554 #e.g. (A or nan --> nan)
|
|
|
555 ignore_nan = False
|
|
|
556
|
|
|
557 #compure ras
|
|
|
558 ras_df,genes_not_mapped=computeRAS(dataset,rules,
|
|
|
559 rules_total_string,
|
|
|
560 or_function=np.sum, # type of operation to do in case of an or expression (max, sum, mean)
|
|
|
561 and_function=np.min,
|
|
|
562 ignore_nan=ignore_nan)
|
|
|
563
|
|
|
564 #save to csv and replace nan with None
|
|
|
565 ras_df.replace([np.nan,None],"None").to_csv(ARGS.ras_output, sep = '\t')
|
|
|
566
|
|
|
567 #report genes not present in the data
|
|
|
568 if len(genes_not_mapped)>0:
|
|
|
569 genes_not_mapped_str=", ".join(genes_not_mapped)
|
|
|
570 utils.logWarning(
|
|
|
571 f"INFO: The following genes are mentioned in the GPR rules but don't appear in the dataset: "+genes_not_mapped_str,
|
|
|
572 ARGS.out_log)
|
|
|
573
|
|
|
574 print("Execution succeeded")
|
|
|
575
|
|
|
576 ###############################################################################
|
|
|
577 if __name__ == "__main__":
|
|
539
|
578 main() |
